Note: If you are interested in leveraging Twitter data for brand monitoring purposes, you should consider attending our 140| Twitter Conference, as we are hosting a session that focuses on this.
As we did with our recent Kindle DX analysis, Warren Sukernek of Radian6 has sent me a data set containing 5,000 of the earliest tweets mentioning “star trek.” These were pruned down to a set of 700 or so that contained text strings likely to indicate the writer of the tweet had actually viewed the film. We then prioritized these by reach and extracted a set of 200 tweets which encompassed 90% of the entire reach/followership of those 700.
We ran the resulting set through Sentimine, and had each tweet tagged for sentiment. In addition we had the system attempt it’s own confirmation of “viewership” — in other words, elimination of records that did not indicate the “Tweeter” had seen the movie.
The resulting group of 144 tagged records/tweets reached 74,000 potential followers and were 86% positive by quantity, 69% by reach. See the charts below.
We’ve never before seen such low “neutrals” combined with very low “negatives.” That being said, we typically evaluate consumer electronics product launches, and movies present a very different case. If you look at numbers presented at rottentomatoes.com, (96% on the Tomatometer) I feel that our numbers align with reality. If I was a Viacom shareholder, I’d be feeling pretty good right now.
As I’ve said before, while our sentiment accuracy is superior to what’s offered via the Twitter API, this data reflects just a most basic pass through the system. It’s likely a couple of sentiment scores need adjusting. If anyone wants the data set to examine, email me and I’ll send it off. Steve AT parnassusgroup DOT com.
Note: Thanks to the great data that Radian6 supplies, we can analyze just about any product/brand for sentiment. If you have a Radian6 account, we can analyze the data you’re receiving too. Contact me for more info.
{ 1 trackback }
{ 0 comments… add one now }